THE BEST OF BUSINESS ECONOMICS HIGHLIGHTS FROM THE FIRST FIFTY YEARS

So long as that is the target, instances seem sure to arise in which the strength of demand pressures will be misjudged; fi scal restraint will be applied too late or not at all; and the

THE BEST OF BUSINESS ECONOMICS

HIGHLIGHTS FROM THE FIRST FIFTY YEARS

Selection and editorial content © National Association for Business Economics and Robert Thomas Crow 2016

Individual chapters © their respective contributors 2016

All rights reserved No reproduction, copy or transmission of this publication may be made without written permission No portion of this publication

may be reproduced, copied or transmitted save with written permission In accordance with the provisions of the Copyright, Designs and Patents Act

1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, Saffron House, 6-10 Kirby Street, London EC1N 8TS.

Any person who does any unauthorized act in relation to this publication

may be liable to criminal prosecution and civil claims for damages.

First published 2016 by

PALGRAVE MACMILLAN

The authors have asserted their rights to be identified as the authors of this work

in accordance with the Copyright, Designs and Patents Act 1988.

Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited, registered in England, company number 785998, of Houndmills, Basingstoke, Hampshire, RG21 6XS

Palgrave Macmillan in the US is a division of Nature America, Inc., One

New York Plaza, Suite 4500, New York, NY 10004-1562.

Palgrave Macmillan is the global academic imprint of the above companies and has companies and representatives throughout the world.

E-PDF ISBN: 978–1–137–57251–6

DOI: 10.1007/978-1-137-57251-6

Distribution in the UK, Europe and the rest of the world is by Palgrave

Macmillan®, a division of Macmillan Publishers Limited, registered in England, company number 785998, of Houndmills, Basingstoke, Hampshire RG21 6XS Library of Congress Cataloging-in-Publication Data

The best of Business economics : highlights from the first fifty years / National Association of Business Economics; edited by Robert Crow.

pages cm

Includes bibliographical references.

1 United States—Economic policy 2 United States—Economic

conditions —1945– 3 Economics—United States 4 Business—United States

5 Managerial economics—United States I Crow, Robert, editor II National Association for Business Economics (U.S.), issuing body III Business economics (Cleveland, Ohio)

Introduction xvii Robert Thomas Crow

1 A New Look at Monetary and Fiscal Policy (1967) 3

Paul A Volcker

Although the recent past amply demonstrates the temptation for policy makers

to discount the complications that today’s actions may create for tomorrow, it would be a serious mistake to confine policy makers to a fixed monetary rule Part of our recent difficulties stemmed from underlying developments in finan-cial markets, such as the liberalization of institutional lending, that weakened the “availability effects” of monetary policy For the longer run, actions are needed to make financial markets more evenly responsive to monetary policy

2 The Role of Money in Economic Activity: Complicated or

3 Econometric Model Building for Growth Projections (1969) 19

Lawrence R Klein

The Wharton Model is extrapolated for 24 quarters with appropriate tions The problems involved in specifying a fresh model for long range forecasting are then outlined, including “endogenizing variables” such as gov-ernment spending Attention is drawn to the less precise performance in pre-dictions of the long run Skepticism is expressed concerning deceptively smooth and free hand extrapolations

4 Presidential Address: The Challenge to Our System 31

Alan Greenspan

Concern is expressed about the extent of government intervention Yet ernment has a role to play, e.g in industrial pollution But there are problems

gov-A law of fiscal constituencies is formulated: the growth rate of benefits to stituent groups tends to exceed the fiscal dividend Grave questions remain about the fiscal outlook in coming years

5 The Social Significance of Environmental Pollution (1970) 39

Barry Commoner

Environmental pollution is not an incidental by-product Rather, it is an sic feature of the very technology developed to enhance productivity This technology is so imbedded in the agricultural and industrial production pro-cesses that the required change would involve serious economic dislocations The author contends that the problem is so serious that these dislocations must

7 Why Productivity Is Important (1973) 59

Geoffrey H Moore

Productivity growth has played a key role in insuring higher real wages and in combating inflation over the last quarter of a century These facts and others relating to productivity are documented as the relation of hourly compensa-tion, productivity and unit labor costs is sketched, and the relation of the latter

to total costs, prices and profits is outlined Future real economic growth out inflation will depend on high rates of productivity growth

inflation, cost or wage push inflation, and shortage inflation stemming from special factors Each type is analyzed and suggestions made as to how it can be treated Special emphasis is placed on cost or wage push inflation and from what causes it arises The spectrum of opinion on the part played by unioniza-tion in wage push inflation is reviewed Finally, various anti-inflation policies are examined and some international aspects of inflation touched upon

10 The Practical Use of Economic Analysis in Investment

Edmund A Mennis

Economic analysis can be most effective if it is fully integrated into the ment decision process Here, the investment decision process is described Specific uses of economic analysis in the various parts of the process are detailed, examples are given, and certain caveats provided

11 Presidential Address : On Human Welfare (1979) 95

Albert G Matamoros

As important as it is continuously to assess our role as business economists, I think it is equally imperative that we step back, from time to time, and examine still broader issues Today I want to raise some questions regarding the extent to which economic policies and the consequent actions of the agencies of gov-ernment during the past 15 years have contributed to the human condition It

is not only appropriate, but I think mandatory, that, as social scientists, we be concerned for man’s welfare and his destiny

12 Company Total Factor Productivity: Refinements, Production

Functions, and Certain Effects of Regulation (1981) 105

Douglas L Cocks

The current concern over the lack of productivity growth in the US mandates certain actions by companies One of these actions is the measurement of pro-ductivity for individual firms This chapter presents some refinements in the measurement of Total Factor Productivity (TFP) at the firm level In addition, alternative methodologies are investigated with the result that these alternatives yield consistent results The chapter also demonstrates two applications of the TFP model to empirical investigations relevant to public policy issues: the impact of regulation on measured productivity and estimation of production functions for the firm One interesting empirical result is that, given the neces-sary input data, the negative effects of regulation on productivity can be dem-onstrated through the TFP model

13 The Adam Smith Address : Conservatives, Economists, and

Herbert Stein

One might expect Adam Smith to be the patron saint of economists of all ideologies He was the father not only of a particular idea of how the economy works but also of the idea that there is an economic system Moreover, some of his ideas about how the system works are incorporated in all kinds of economics, from extreme left to extreme right Any economist teaching the history of economic thought would start with Adam Smith But the wearing of Adam Smith neckties is not uniformly or randomly distributed among economists Only economists who are, loosely, called conservatives wear it

14 Economics from Three Perspectives (1982) 129

Marina v N Whitman

This chapter is based on a talk Dr Whitman gave at Notre Dame University In

it, she shares some of her personal views on the various roles economists play

in society, the need for greater interaction among academic, government and business economists and the evolution of the role of corporate economists at General Motors Corporation

15 The Adam Smith Address : Was Adam Smith a Monetarist or

a Keynesian? (1984) 137

Charles P Kindleberger

I give this talk the foregoing title (a) because it is the Adam Smith Lecture and (b) because I want to hold forth on Keynesianism and monetarism It is evi-dent, however, that to put the matter as a choice between a single pair of alter-natives is fallacious Adam Smith was and is under no compulsion to fall exclusively into one category or the other

16 The Adam Smith Address : The Effect of Government on

Economic Efficiency (1987) 151

George J Stigler

This chapter examines the effects of governmental policies upon the efficiency

of the economy, including both traditional governmental areas (such as justice, defense, and environmental protection) and the vast and growing share of gov-ernmental programs aiming to redistribute income A proposed principle of legitimacy states that every action set by a legislature represents a social judg-ment that society is better off for that action Thus all governmental policies are

by hypothesis utility-increasing for the nation Any costs of (say) a tion of income are less than the benefits National output as presently measured can and usually will fall when a new redistribution of income is instituted, because it is costly to redistribute income Is this trend in governmental policy likely to be reversed, perhaps by a general movement toward deregulation? The author’s answer to this question is calculated to restore the claim that econom-ics is the dismal science

17 The Adam Smith Address : On the Structure of an Economy (1988) 161

or results of market process within structures are based on fundamental derstanding

18 Rethinking International Trade (1988) 171

theory may strengthen the arguments for free trade, it also alters recommended government trade policy Subsidies may tilt competition in favor of a high return domestic industry, giving it a head start and a persistent advantage While this new trade theory may not always be effective, it does change free trade from a dogma to a reasonable rule of thumb in an imperfect world

19 The Adam Smith Address : The Suicidal Impulse of the Business

20 A Guide to What Is Known about Business Cycles (1990) 189

Victor Zarnowitz

This chapter reviews the common core of the pervasive and persistent sonal fluctuations that have characterized modern capitalist economies But much diversity also exists, and the differences between cycles before and after World War II are discussed Some reasons for these changes are offered Finally,

nonsea-a brief comment considers the vnonsea-arious theories nonsea-advnonsea-anced to explnonsea-ain “the” cycle, and the difficulty of so doing because cycles are not all alike

21 Some Financial Perspectives on Comparative Costs of Capital (1991) 203

J Fred Weston

Empirical studies of international cost of capital comparisons have taken two related forms One is to compare weighted average costs of capital (WACC) for samples across economies Sample WACC comparisons may be subject to error because the cost of capital measures may not be applied to appropriate defini-tions of operating cash flows whose qualities, time-growth patterns, and risk may differ Comparisons of riskless rates such as yields on government securi-ties ignore relevant risk differences No financially derived competitive advan-tage is likely to exist with: (1) no net tax or subsidy differences, (2) capital market and economic integration

22 Health Insurance Derivatives: The Newest Application of Modern

Financial Risk Management (1993) 213

James A Hayes, Joseph B Cole , and David I Meiselman

This chapter discusses the derivatives revolution in financial and other markets, emphasizing the gains in market efficiency and innovation by reducing transac-tion costs and promoting new product development Health insurance futures and options, a natural extension of the derivatives revolution, will be trading at the Chicago Board of Trade in 1993 In addition to an overview of the struc-ture of the health insurance futures contract, an example is given of a long hedge by an insurance company to protect itself from unexpectedly higher claims payments

24 The Adam Smith Address : Capitalism and Its Discontents (1998) 231

of government, needs its defenders, as the alternative models have proven torically, intellectually and practically bankrupt

25 Protecting Against the Next Financial Crisis: The Need to

Reform Global Financial Oversight, the IMF, and Monetary Policy

Henry Kaufman

Recent distress in world financial markets has underlined the need for vising and regulating financial institutions and markets on a global basis A new institution, in addition to the International Monetary Fund (IMF) and the World Bank, is required to set forth a code of conduct to encourage reasonable financial behavior and to supervise risk-taking It also should be empowered by member governments to harmonize minimum capital requirements, to estab-lish uniform trading, reporting and disclosure standards, and to monitor the performance of institutions and markets under its purview The IMF should be able to demand policy changes in anticipation of problems Securitization and the development of financial derivatives have liberalized granting of credit, requiring steeper interest rates to end a period of excessive monetary expan-sion Monetary policy also should be concerned with asset inflation as well as price inflation

26 How the Economy Came to Resemble the Model (1999) 255

Alan S Blinder

Over the years, economists have spent much effort to modify the capitalist, perfect-competition, profit-maximizing model of classical microeconomics to fit reality Thus, it is ironic that in recent times reality has been approaching the classical model This is due only in part to the persuasive talents of economists, and not all of this change is necessarily an improvement Among the factors contributing to the reversion to the classical model are the failure of socialism, alignment of managerial and shareholder interests, focus on shareholder value, decline of labor union power, changes in financial markets, global competition, and changes in regulatory practices

27 The Adam Smith Address : What Would Adam Smith Say Now? (2000) 271

Henry Kaufman

The breadth and depth of Adam Smith’s thought over 200 years ago still vide powerful lessons today Were he present, he would applaud much of what has transpired in the organization of economic life, particularly in the US economy and its thrust toward individual achievement and relatively free mar-kets for goods and services, capital, and labor However, he would also be deeply troubled by recent trends toward consolidation, particularly in the financial sector, and the emergence of “too-big-to-fail” as an argument for government

pro-to weaken the discipline of markets

28 Information Technology and the U.S Productivity Revival:

A Review of the Evidence (2001) 279

Kevin J Stiroh

Aggregate, industry, and firm level studies all point to a strong connection between information technology (IT) and the US productivity revival in the late 1990s At the aggregate level, growth accounting studies show a large and growing contribution to productivity growth from both the production and the use of IT At the industry level, industries that produce or use IT most intensively have shown the largest increases in productivity growth after 1995

At the firm level, IT-intensive firms show better performance than their peers, and several specific case studies show how IT improves real business practices This accumulation of evidence from a variety of studies suggests a real produc-tivity impact from IT

29 Presidential Address : Understanding Inflation: Lessons from My

Central Banking Career (2002) 291

Harvey Rosenblum

Economic theory—much less modeling based on historical data—has a cult time keeping up with structural change in the contemporary economy Anecdotal evidence and a feel for the economy based on experience are likely

diffi-to be as important as theory-based modeling in making real-time policy sions on the control of inflation and the stability of the economy Many of the phenomena to be understood are microeconomic in nature While much has been learned about effective stabilization policy over the past forty years, econ-omists still have a long way to go before inflation can be understood and man-aged

30 Managing Exchange Rates: Achievement of Global Re-Balancing or Evidence of Global Co-Dependency? (2004) 305

Catherine L Mann

Long-term global economic health requires that external imbalances and the internal imbalances that support them be corrected by both the United States and its trading partners The current path of external imbalances appears to be unsustainable, but relying on exchange rate adjustments is unlikely to suffice as long as there is a co-dependency of structural characteristics and policy choices between the United States and its trading partners There is a real possibility that the entanglements created by this co-dependency cannot be undone by anything short of a global economic crisis

31 The Adam Smith Address : The Explanatory Power of Monetary

Policy Rules (2007) 321

John B Taylor

Over the past 20 years, the use of monetary policy rules has become pervasive

in analyzing and prescribing monetary policy This chapter traces the ment of such rules and their use in the analysis, prediction, and stabilization of national economies In particular, rules provide insight into eras in which mon-etary policy was not effective as well as when it was, such as the persistence of the ongoing “Great Moderation.” The chapter stresses “the scientific” contri-butions of rules, including their insight into fluctuations of housing construc-tion and exchange rates, as well as into the term structure of interest rates

32 The Adam Smith Address : Adam Smith and the Political Economy of a

Modern Financial Crisis (2008) 335

33 Underwriting, Mortgage Lending, and House Prices:

James A Wilcox

Lowering of underwriting standards may have contributed much to the unprecedented recent rise and subsequent fall of mortgage volumes and house prices Conventional data do not satisfactorily measure aggregate underwriting standards over the past decade: the easing and then tightening of underwriting, inside and especially outside of banks, was likely much more extensive than

they indicate Given mortgage market developments since the mid-1990s, the method of principal components produces a superior indicator of mortgage underwriting standards We show that the resulting indicator better fits the var-iation over time in the laxity and tightness of underwriting Based on a vector auto-regression, we then show how conditions affected underwriting stan-dards The results also show that our new indicator of underwriting helps account for the behavior of mortgage volumes, house prices, and gross domes-tic product during the recent boom in mortgage and housing markets

34 The Impact of the Housing Market Boom and Bust on

Consumption Spending (2010) 377

Jeremy A Leonard

While econometric evidence for the United States has consistently shown that increases in real estate wealth induce additional consumption, it does not directly speak to the effect of a substantial decrease in real estate wealth This chapter examines the real estate wealth-consumption relationship over the past half century with a particular focus on the sharp decline in 2006–2008, and finds that the wealth effect in the recent down market is significantly larger than in an up market Additionally, wealth changes seem only to affect con-sumption of services and nondurable goods; there is virtually no impact on durable goods consumption

35 The Adam Smith Address : Macroprudential Supervision and Monetary

Policy in the Post-Crisis World (2010) 393

be done to identify emerging systemic financial risk, the tools and tation of macroprudential financial supervision that must be developed, and the role of coordination between monetary policy and macroprudential super-vision Prevention of crises will not be easy—particularly because it will be necessary to walk a tightrope between prevention of catastrophe and keeping too tight a hold on the financial system

36 The Adam Smith Address : Nightmare on Kaiserstrasse (2011) 407

37 The Adam Smith Address : Financial Services and the Trust Deficit:

Why the Industry Should Make Better Governance a Top

corpo-so important that it is critical that trust be restored This is particularly true in

an era in which planning for retirement is inadequate, and financial literacy is increasingly important This implies a critical role for a financial services indus-try whose guidance can be trusted This trust must be built through appropriate corporate governance on the part of all stakeholders, and specific recommen-dations are advanced for bringing this about

38 US Economic Prospects: Secular Stagnation, Hysteresis, and the

Zero Lower Bound (2014) 421

Lawrence H Summers

The nature of macroeconomics has changed dramatically in the last seven years Now, instead of being concerned with minor adjustments to stabilize about a given trend, concern is focused on avoiding secular stagnation Much

of this concern arises from the long-run effects of short-run developments and the inability of monetary policy to accomplish much more when interest rates have already reached their lower bound This address analyzes contemporary macroeconomic problems and proposes solutions to put the US economy back

on a path toward healthy growth

39 Focus on Industries and Markets: Electric Power Transmission and Distribution

David A Petina, Michael Murphy, and Andrew C Gross

The US electrical grid must be upgraded, and there is a strong debate about the characteristics of the next-generation electrical network However, slow growth of electricity usage, among other factors, means that the demand for transmission and distribution (T&D) equipment is growing slowly also Within the T&D equipment sector, switchgear and transformers are still the dominant segments, but sales of meters are growing rapidly in response to increased demands for security, safety, and connectivity Six firms hold about 40 percent

of the T&D equipment market share, selling to electric utilities, nonutility industrial firms, commercial firms, and residential customers Foreign trade is also important in this industry, with the United States running a substantial trade deficit

40 Focus on Statistics: Initial Results of the 2012 Economic Census (2014) 455

Robert P Parker

The initial results of the 2012 Economic Census are being released and will continue to be released until mid-2016 It provides detailed industry and geo-graphic data used by businesses, researchers, and government policy makers It

provides the detailed data used to benchmark the Index of Industrial Production and the Producer Price Index, and to prepare input-output accounts and quar-terly GDP The Census Bureau uses the data to benchmark most of the Census Bureau’s annual, quarterly, and monthly economic surveys and this bench-marking maintains the reliability of these sample surveys The 2012 Economic Census is essentially the same as 2007 census

41 Economics at Work: Economics at the American Chemistry

Thomas Kevin Swift

This chapter examines the role of the economics team at the American Chemistry Council, a major trade association representing the leading compa-nies in the business of chemistry The history of the team, its organization, its role in providing good statistics, monitoring and forecasting business condi-tions, conducting policy analysis, and thoughts on managing professionals are presented

“If all the economists were laid end to end, they’d never reach a conclusion.”

George Bernard Shaw

In 1959, the National Association of Business Economists—now the National Association for Business Economics (NABE)—was formed In 1965, its

Governing Council decided to create a professional journal, Business Economics This collection of articles commemorates Business Economics’ 50th anniversary It is

an attempt to pull together the best from each decade of those 50 years

Business Economics was intended by its founders as a professional journal, not an

academic one Throughout its history, although editorial policies have changed from time to time, this intent has been consistent 1 Its focus has been on what is useful to economists working in business rather than on advancing economic sci-

ence Often, however, an article will be useful and advance economic science—so much the better Excerpts from Business Economics’ current instructions to authors

refl ect the journal’s consistent intent:

“To be considered for publication, there must be a clear statement of how and why the analysis and information presented in the paper are important to business That is, articles must pass the ‘So what?’ test.”

“ Business Economics readers are professionals, managers, and researchers in

organiza-tions of all sizes and kinds, primarily in North America Their interests—improving their performance on the job and using economics in the workplace—shape what

Business Economics editors look for in reviewing submissions.”

A horseback guestimate is that about 1,200 regular articles have appeared in Business

Economics over its 50-year history This implies an immense job of screening and

eval-uation to select the 3 percent or so that will actually be published To do this, the Editorial Board and feature editors were recruited into fi ve teams—one per decade—

to select the best articles However, Business Economics over the years has some traits that

permitted some screening so that the teams would not have to review every article

In 1986, NABE inaugurated the Abramson Award for the outstanding paper

published in Business Economics in the preceding year That year, it also inaugurated

the Adam Smith Award for “leadership in the profession and the application of economic principles and knowledge in the workplace and policy arenas.” Upon receiving this award, the recipient gives an address that is subsequently published

in Business Economics Thus, for 1986 to 1999, only the President’s Address articles,

the Adam Smith Award articles, and the Abramson Award articles were considered

In 1999, winners of the competition for the Mennis Award were added; and in

2014 NABE inaugurated the Lifetime Achievement Award for Economic Policy, the presentation of which is accompanied by an address by the recipient These awards provided a means of screening the articles to be considered Although some outstanding articles that have proven to be more valuable through the test of time than they appeared to be initially may have been lost in this process, it seemed like

a reasonable way to make selections

However, “best” in economics—as in art—is a slippery concept In what sense is Bach better than Beethoven, or vice versa? What about Mozart and Tchaikovsky? Hence, the George Bernard Shaw quote above Few articles were consensus choices; most were by plurality Finding a consensus on the best economics articles is not unlike corralling a hatful of mice

Even though it was possible to limit the population from which the selections were made, selecting was still a demanding, time-consuming job, and I would like

to acknowledge those who took on this task They are as follows:

Although the content of Business Economics is quite varied, the overwhelming

number of the articles selected for this volume concern the environment in which

fi rms do business rather than the role of economics within the fi rm or the how nomics should be practiced in business The latter concern is often addressed in a fea-ture entitled “The Business Economist at Work” or “Economics at Work” that appears

eco-in many issues Other features are “Focus on Statistics” and “Focus on Industries and Markets.” One example of each of these features is included in this volume

There is a great deal of wisdom and inspiration in the articles that have been selected Some are uniquely relevant to the time in which they were written Most, however, still speak to economists and those who use economics in their work I hope that this volume will be a source of information and inspiration to these readers

Robert Thomas Crow

Editor, Business Economics

2015

Note

1 An article on the history of Business Economics appears in the October 2015 issue

Roger C Bird Lynn O Michaelis

Oral Capps Gerald L Musgrave

Glenn R DeSouza Francis H Schott

William C Dunkelberg Nancy D Sidhu

Robert A Eisenbeis Thomas F Siems

Elinda Fishman Kiss James F Smith

Andrew C Gross Charles Steindel

J Paul Horne Christopher M Swann

Parul Jain T Kevin Swift

Douglas J Lamdin Diane C Swonk

1965–1974

1967

A NEW LOOK AT MONETARY AND FISCAL POLICY

Paul A Volcker, Chase Manhattan Bank

The simple fact of the matter is that I cannot detect—out of all the welter of commentary on 1966—an agreed doctrine emerging on the delicate job of managing prosperity

There is, I think, a wide consensus on two basic points

First, nearly everyone agrees that fiscal policy needs to play a more effective

•

role than has been evident in the past two years

Second, the point has been well made—particularly by one of NABE’s past

Hindsight vs Foresight

I agree with the consensus on those points But I fi nd them of decidedly limited usefulness in terms of practical guidance for the policy-maker in business or gov-ernment Take the fi scal policy question In retrospect, it is amply clear that taxes should have been raised in early 1966 But at the time, I was not exactly deafened

by the pleas of professional economists for action along those lines

That contrast between hindsight and foresight—even when the basic economic principles are widely understood—is hardly surprising to a group of business econ-omists We deal regularly with the uncertainties surrounding business decision-making in a complex world—and we are well aware that there are, quite properly, more ingredients in nearly all decisions than a question of economic judgment I would only point out that the political setting is still more diffi cult, particularly in

so sensitive an area as tax and expenditure policy, and we cannot assume that fi scal policy will always be unerringly in tune with our needs

I do not mean to be defeatist The fact that the Administration is pushing hard for

a tax increase now, and with a great deal of professional and business support, seems

to me evidence of progress In particular, I believe the President and his advisers are quite right in recognizing that to wait until all the evidence is in—to wait until the business expansion has already reached the point of adding further fuel to the infl ationary pressures—would be to wait far too long Yet, clear as the need seems

to me, prospects for action are obscured by the entirely legitimate related questions

of spending policy, as well as by the doubts as to whether tax action is premature,

or excessive or necessary at all

The diffi culties of fi ne tuning fi scal policy are obviously related to forecasting problems Something in the mystique that still surrounds central banking—plus some real elements of greater fl exibility—have made monetary policy less vulner-able to the charge of resting on weak forecasts But the lags in policy impact are clearly there So long as they are, so is the forecasting problem

The combination of policy lags and forecasting errors assure that it will always

be possible, in looking at the past record, to fi nd specifi c instances of when it would have been better to have done nothing—or to have followed a fi xed rule—than to have done what was in fact done Plenty of examples can be found in the past two years And these examples have, I suspect, provided a more sympathetic audience for those who favor a mixed rule, whether of the monetary or the fi scal variety

I would myself concede that the recent period has amply demonstrated the strong temptation for policy makers—in their impatience to deal with today’s

fi res—to underestimate the lags in policy transmission, and to discount too ily the possible complications that today’s action is creating for tomorrow But I remain a very long way from wanting to put blinders on the policy makers by confi ning them to a fi xed monetary policy rule In fact, recent experience provides

heav-an exceptionally clear demonstration of how violently liquidity preferences may shift in response to expectations or other factors, and how quickly and massively an established process of fi nancial intermediation can be distorted These seem to me precisely the kind of economic phenomena that cannot be dealt with satisfactorily

within the framework of some a priori judgment as to a desirable rate of increase in

some single monetary or credit variable

poli-no set formula for managing prosperity, and poli-no guarantee against a repetition of problems in fi nancial markets akin to those of 1966

The situation last year was abnormal in one respect The economy is rarely called upon to absorb so rapid—and so poorly estimated—a military buildup But in

another, and even more fundamental, respect it was presumably not so abnormal: a basic aim of economic policy is clearly to make full employment the norm So long

as that is the target, instances seem sure to arise in which the strength of demand pressures will be misjudged; fi scal restraint will be applied too late or not at all; and the monetary authorities will be left with a sharp dilemma, seemingly caught between the twin evils of underwriting infl ation through excessive doses of credit

or of pursuing restraint to the point of demoralizing capital markets, with distorting and unpredictable repercussions for the real economy

The seeming absence of much middle ground between those extremes in 1966—and potentially again in 1968—is partly a refl ection of the degree to which

fi scal policy is out of tune But I suspect that some underlying developments in

fi nancial markets over a period of years also help to account for the sharpness of the dilemma Without attempting to list or analyze the relevant changes in detail, I would point to such developments as the negotiable CD and the Eurodollar market, more aggressive and imaginative competition for other varieties of time money and savings among banking and savings institutions, the relatively liberal administration

of offi cial interest-rate ceilings on bank deposits during the fi rst half of the 1960s, a greater fl exibility in portfolio policies by institutional lenders and liberalization of self-imposed or offi cial restrictions on institutional lending policies The net eff ect was to increase markedly the elasticity and fl uidity of credit markets A more perfect market was created, in the sense that changes in demand and supply were refl ected more fully through interest rates and less through changes in credit availability

An economist is inclined to look upon improved market performance as an unmitigated blessing During the early 1960s, the kinds of developments to which

I am referring contributed very signifi cantly to the growth and balance of the economy But from the viewpoint of those called upon to apply a restrictive credit policy, a more perfect market also generated drawbacks

In particular, the early moves toward restraint were refl ected more in higher interest rates—and less in reduced availability—than seemed either politically

or economically tolerable The economic problem arose because of the relative inelasticity of spending with respect to interest rates—at least in the short run and in those sectors of the economy that were the source of greatest concern I

fi nd no evidence from recent experience to cast doubt on the prevailing view that it is the availability eff ects of monetary policy that count most heavily in terms of eff ective short-run restraint on business With these availability eff ects weakened, the real economy had become less quickly and predictably respon-sible to traditional Federal Reserve controls operating primarily through bank reserves

The obvious answer seemed to be to push restrictive policies harder, at the expense of still higher rates And in time this approach did expose some very basic institutional rigidities Many savings institutions, dependent on borrowing short but locked into long-term assets acquired earlier at lower rates, were simply unable

to keep pace with the increases in market rates Disintermediation found its way into our vocabulary, and this indeed triggered strong availability eff ects

The trouble was that the appropriate dosage of this strong medicine proved diffi cult to prescribe and control, and the side eff ects were serious The resulting

dislocations and imbalances in the credit markets quickly fed back into the real economy

In attempting to even out the impact on fi nancial markets and moderate the rate pressures, while maintaining eff ective restraint, the authorities manipulated what instruments of selective control they had readily at hand—notably interest-rate ceilings, selective adjustments in reserve requirements, and by September 1966, the more direct approach of spelling out for the banks what type and how much lend-ing was appropriate

In somewhat cavalier fashion, I would sum up this episode as a piecemeal and unplanned eff ort to improve the linkages between Federal Reserve action and economic reaction Intense restraint was certainly achieved But I think it also fair

to conclude that, in pushing the credit and capital markets close to the point of paralysis, eff ective control over the degree of restraint was lost

Edge of Urgency

I have reviewed this familiar ground in some detail because I believe a very similar dilemma for monetary policy could arise again, and the authorities seem little more equipped to deal with it In fact, there are serious complicating factors growing directly out of the experience a year ago Infl ationary expectations and real cost pressures are more solidly entrenched Lenders and borrowers alike have under-standably wanted to restore liquidity to insulate themselves as best they can from a repetition of 1966; faced with a business slowdown, the Fed had little choice but to meet those liquidity desires through a fresh outpouring of bank credit Prolongation

of that process as business picks up has obvious infl ationary dangers But how can the process be stopped when there is a ballooning defi cit to be fi nanced, when the market is hyper-sensitive to any restraint, and when memories of the near chaos of August 1966 are fresh in mind?

All this is what puts the edge of urgency on the current proposal to raise taxes

It is also this prospect that seems to be leading some money market observers to

a conclusion that we must, almost inexorably, move toward direct credit controls

if the tax bill fails, and maybe if it doesn’t The argument is insidiously simple Monetary restraint, applied strongly enough to be eff ective in the short run, will risk a repeat of the 1966 distortions and crisis atmosphere The kind of ad hoc selective measures taken in 1966 provide no protection But, on the other hand, we cannot permit infl ation to be unchecked Ergo, we need to develop a more eff ective system of direct controls

This is not the place for a philosophical discussion about the inadequacies of direct controls More immediately to the point, there are some practical consider-ations that cast into doubt the usefulness of direct controls as a means of meeting our problems

We had a taste last year of the way controls, applied piecemeal and selectively, can distort markets and have perverse eff ects on expectations and psychology We

do not have the kind of clear national emergency that in the past has been the only justifi cation for more comprehensive controls And, even if these powers did exist, it

is not apparent how an apparatus of controls designed to restrain the private market could redress an imbalance so clearly arising in the federal sector

I earlier indicated my belief that the kind of problem we faced in 1966 will not prove unique Partly for that reason, I would resist the thought that “controls are all right—after all we are in a war.” And I feel certain that the policy makers in Washington do not want to see the great experiment of the New Economics give way to a network of direct controls

Limits to Monetary and Fiscal Policy

For the longer run, all of us could list many specifi c actions that might improve the performances of fi nancial markets—and at the same time make them more evenly and predictably responsive to monetary policy My personal list would include such broad and complex matters as measures to improve the liquidity of the mortgage market—and to strengthen the weaker links in our structure of sav-ings institutions

I would also include a new look at the Federal Reserve discount window in an eff ort to break down what seems some overly rigid attitudes on the part of both member banks and the Federal Reserve concerning its use—a matter under close offi cial examination I would certainly undertake a careful reappraisal of the role that interest rate ceilings should play—if any at all I would hope, too, that private lenders and borrowers have learned to maintain better control over their forward commitments

But most of all I believe the fundamental lesson of the past two years has been to reemphasize what we already knew: there are limitations on the ability of monetary and fi scal policy to keep the economy moving ahead at a steady and fully employed pace I do not want to underestimate the real achievements of the past two years But I believe that they also illustrate that we have not yet learned how to reconcile full employment with price stability And that is a diffi culty that arises more in labor markets than in money markets

Note

Originally published in Business Economics , Vol 3, No 1 (Fall 1967), pp 29–31

1969

THE ROLE OF MONEY IN ECONOMIC ACTIVITY:

COMPLICATED OR SIMPLE?

Edward M Gramlich, Board of Governors of the

Federal Reserve System

A hot dispute currently rages as to the importance of money in influencing economic activity But we would be wrong to think this a new controversy Indeed, it is a very old controversy which has been with us for decades Money was all-important in classical models of the economy but much less so in Keynesian models which gained predominance in the Great Depression and continued into the postwar period Lately, however, there has been a strong revival of interest in monetary phenomena and this revival has led to the current heated dispute on the importance of money

The main reason for diff erences of opinion on the importance of money has been the diffi culty in obtaining convincing empirical evidence concerning the sensitivity of aggregate demand to monetary and fi scal forces Historical evidence suggests that autonomous monetary forces such as gold discoveries and reserve requirement decisions played a major role in the infl ation of 1900–1910, the Great Depression, and the contraction of 1936–1937 These fi ndings are buttressed by the one equation studies of Friedman-Meiselman, Andersen-Jordan, and others of the monetarist persuasion, which have invariably found monetary, variables to be much more important than fi scal variables in explaining subsequent movements in GNP

On the other hand, the evidence from the large econometric models—the Wharton School model, the OBE model, the Michigan model, and the Brookings model—is that monetary forces are rather unimportant in infl uencing total demand

The FRB-MIT econometric model originated in this controversy The ject has been under the joint direction of Frank de Leeuw at the Board and Professors Franco Modigliani of MIT and Albert Ando at Penn, the latter two of whom were spurred on in an attempt to resolve their inconclusive interchange with

pro-Friedman-Meiselman in the 1965 American Economic Review The aim of the project

was to build a model which, though not necessarily larger than most other ing models, would focus more intensively on monetary forces and the way they aff ect the economy We used an econometric model because we wanted above all

exist-to explain the structure of the relationships involved in the transmission of etary infl uences to the real economy We felt that the US economy is so complex, with such a large number of important relationships, so many diff erent monetary and fi scal policy instruments having such diverse eff ects and such complicated and variable time lags, that a simple, cheap one equation approach could not possibly

mon-do justice to the problem 1

Three Channels

A basic diff erence between our model and other large-scale econometric models

is that we have incorporated additional channels through which monetary forces aff ect economic activity All models have what we call a cost-of-capital channel,

by which interest rates aff ect investment in real capital In our model this capital infl uence works on plant and equipment investment, on housing, on invest-ment in consumer durables, and on the construction expenditures of state and local governments No other model includes the latter infl uence, and most other models

cost-of-do not spell out the other three cost-of-capital infl uences as completely and sistently as our model does Thus even for the cost-of-capital channel we think our model makes a considerable advance over other models in determining the quanti-tative importance of monetary forces I might add that our model shows a stronger monetary infl uence through the cost-of-capital channel than other models do

A second channel of transmission of monetary forces is through the net worth of consumers An important link in this mechanism is the stock market: interest rates

on long-term bonds infl uence the rate at which the stock market capitalizes dend payments; this dividend-price ratio determines the value of common stock in net worth; and net worth is one of the factors which infl uences consumption The wealth channel works very much the way Professor Milton Friedman has described monetary policy as working It is very powerful in both the short and long run, and we think it makes great progress in reducing the monetary controversy to one where we can argue about the specifi c values of coeffi cients rather than about the basic structure of macroeconomic systems For these reasons, we think that fi nding this channel to be a signifi cant vehicle for transmitting monetary infl uences has been another important contribution of our model

A third channel of monetary transmission is credit rationing Rationing may be defi ned to include all cases where interest rates alone do not clear fi nancial markets, such that lenders are forced to equate demand and supply by rationing funds By all odds, the most prominent example of this rationing in the present day US econ-omy is in the mortgage market, which is the only rationing eff ect we have had any success in identifying The combination of sluggish deposit rates at thrift institu-tions and constraints on the lending behavior of these institutions means that when market interest rates rise, deposits fl ow out of thrift institutions, force mortgage credit rationing, and depress housing starts It is likely that there are other traces of

credit rationing in our economy, possibly working through commercial loans and inventory investment, but as yet we have not been able to confi rm the importance

of these other types of rationing in aff ecting fi nal demand To the extent our ing channel is important, however, it again reduces the debate to specifi c values of coeffi cients rather than the structure of macro systems

When we simulate our model to determine policy multipliers, we fi nd etary infl uences to be appreciably more powerful in infl uencing fi nal demand than

mon-is found by other large-scale econometric models By way of illustration, the year GNP multiplier for a maintained change in bank unborrowed reserves (imple-mented by open-market operations) is very small for the Michigan model, about

three-3 for the Wharton School model, about 8 for the Brookings model, and about 20 for our model This is true even though the corresponding three-year GNP mul-tipliers for government expenditures in the various models are quite similar: 2.5 for the Michigan model, 2.9 for the Wharton model, 2.7 for the Brookings model, and 2.4 for our model Thus money has a powerful infl uence in our model, both absolutely and in relation to the fi scal infl uence Our conclusions are still not clas-sical because we fi nd the intermediate run eff ects of monetary policy to be less than those predicted by a simple quantity theory of money and because we fi nd signifi cant intermediate run eff ects of fi scal policy But, the conclusions are at least more classical than the results of other large-scale models—a fi nding which should please monetarists

There is another way in which our model reaches conclusions pleasing to at least some monetarists We fi nd the lags on monetary policy to be long and variable

Of the three-year multiplier for open-market operations referred to above, a lation done for recent years indicates that only 5 percent of the response in GNP

simu-is achieved in the fi rst quarter, 10 percent by the second quarter, 20 percent by the third quarter, and 30 percent by the fourth quarter It is not until the second year following the policy change that the major impact of these policies is felt Although these long lags do not make it impossible to use monetary policy as a stabilization device, especially against persistent cyclical swings such as the Vietnam War buildup period, neither do they imply that it will be easy to “fi ne-tune” the economy by monetary operations

The Long Lags

There are many reasons for the long lags The fi rst is that we fi nd investment and consumption responding to long-term rates of interest rather than short-term rates Plant and equipment expenditures respond to the corporate bond rate and stock prices, housing responds to the mortgage rate, consumer durables respond to a proxy for the consumer credit rate, and state and local construction responds to the state and local rate Thus even though the Fed can bring about prompt changes in bill rates and free reserves, it takes much longer to infl uence the rates that actually determine investment because of the sluggishness of these long-term rates The second reason for the long monetary lag is the decision lag of investors

It takes fi rms time to order and purchase capital goods once interest rates have changed, it takes homeowners time to alter house purchases once mortgage rates

have changed, it takes state and local governments time to fl oat bonds and build schools once state and local rates have changed Off setting these delays are the infl uences of credit rationing generated by short-run disequilibria in fi nancial mar-kets If the credit rationing channel were more important in our model, we would

fi nd monetary infl uences to operate more quickly

The lags for monetary operations are variable as well as long The main cause of the variability lies in the nonlinear price-wage sector, according to which a given change in real output and employment demand will have much less of an impact

on prices if the economy is at low levels of capacity utilization and high levels of unemployment than if the economy is in a tight capacity range In addition, initial order backlogs, corporate profi ts, capital stocks, interest rates, stock prices, and tax rates all work in one way or another to make the monetary lag variable This lag variability does not necessarily make monetary infl uences less predictable, for all initial levels and tax rates would be known in advance However, it does conform

to the intuitive feelings of most people that in an economy as complex as ours,

it would be most surprising if responses to monetary policy followed a constant schedule every time

Just as we feel that our project has narrowed the monetary debate by showing that monetary forces can be important in a large-scale econometric model, an analytical technique the monetarists generally do not favor, the study by Leonall

Andersen and Jerry Jordan in last November’s St Louis Bank Review has made

numerous statistical improvements over previous correlation studies purporting to show the exclusive infl uence of money on economic activity Instead of using cor-relation coeffi cients and discrete lags, Andersen-Jordan have used multiple regres-sions with fl exible distributed lags Instead of representing fi scal policy by actual government expenditures and receipts, Andersen-Jordan have adjusted these totals for cyclical movements in real income and prices so as to remove possible simul-taneous equations bias which could work against fi scal policy 2 And instead of using money as the independent monetary variable, Andersen-Jordan have gone some way toward making monetary infl uences more exogenous by using the monetary base

Fiscal Policy Still Important

Thus, although for reasons I will indicate below I am not basically in thy with the one equation approach, the Andersen-Jordan results are interesting and suggestive and deserving of careful attention Unfortunately for those of us who believe either in fi scal policy or the quantity theory of money, however, Andersen-Jordan still fi nd that fi scal policy has virtually no eff ect on economic activity and that monetary policy has impacts about twice as great as would be predicted by a simple quantity theory Just to spell this out more clearly, the St Louis model is asserting that as long as the central bank did not allow the money supply to change, the Vietnam War, the 1964 tax cut, the surtax, and the invest-ment credit all would have had no eff ect on aggregate demand At the same time, simple open market operations unaccompanied by fi scal policy have remarkably powerful eff ects

The advantage of a one equation technique such as Andersen-Jordan’s is that if the economy really is too complicated to understand, it may be useful simply to compare inputs such as money and outputs such as GNP It could be, for example, that monetary infl uences are so mysterious and elusive that they cannot be captured even by complicated econometric models Monetary infl uences may work through channels model-builders have not yet discovered or been able to estimate, or they may work on some components of fi nal demand sometimes and other components other times Either way, the eff ect of money on fi nal output may be more regular and predictable than its eff ect on any component of fi nal output We might wonder

at this juncture, however, why the monetary infl uences that are so mysterious and elusive as to defy econometric model-builders would follow GNP with a constant and short lag, which is what Andersen-Jordan have found in their statistical results

Some Drawbacks

At the same time, there are several disadvantages to the Andersen-Jordan one tion approach as it compares to econometric models The fi rst objection has been alluded to already—the inherent simplicity of the one equation model Everybody would agree that our economy is much more complicated than one equation, as indeed it is more complicated than one hundred equations Any model, whether one or one hundred equations, necessarily simplifi es things by aggregating behav-ioral units, omitting variables, including error terms, and other devices The appro-priate degree of disaggregation to strive for in model-building is at bottom a matter

equa-of taste in which the benefi t equa-of increasing the structural richness equa-of the model

is traded off against the cost of increasing the complexities of the relationships, including more mysterious feedbacks, increasing the problems of data management, and being less up-to-date Thus I will not make the unfounded accusation that a one equation model is too simple to use—many people have a justifi able prefer-ence for simple models—but we should recognize that the one equation approach foregoes an awful lot of structural richness

To get some idea of the specifi c nature of this objection, I will state some of

the propositions that are assumed to be true in the Andersen-Jordan model Since

GNP is a function of the adjusted monetary base, the full employment surplus, and

a time trend only, fl uctuations in all other exogenous forces are assumed to have only random eff ects on the economy Fluctuations in exports, strikes, changes in initial stocks of capital and debt, demographic movements, changes in foreign or raw materials, prices, and changes in defense orders all have only random eff ects on aggregate demand All do have eff ects in our model

Since total Federal expenditures were used as an independent variable, the composition of these expenditures between defense purchases, compensation of employees, interest payments, unemployment benefi ts, social security payments, grants-in-aid to state and local governments, foreign aid payments, and post offi ce subsidies did not matter All of these types of spending are assumed to have exactly the same eff ects All have diff erent eff ects in our model

Similarly, indirect taxes, which may even raise current dollar GNP to the extent that they cause higher prices, are treated the same as personal taxes and corporate

taxes, which would certainly lower GNP A corporate tax decrease is treated the same whether it is brought about by rate reductions, the investment credit, or accelerated depreciation provisions 3 Our model treats all of these tax provisions diff erently

On the monetary side, since we are dealing with the monetary base adjusted for liberated reserves, we must assume that a given adjusted base has the same impact

on GNP whether reserve requirements are at present levels or twice as high We must assume that changes in ceiling rates on time deposits, discount rates, the intro-duction of certifi cates of deposits, and rate diff erentials between thrift institution deposits and market instruments all have had no eff ect on the real economy We do not need to make these assumptions in our model In each of these and many other areas, an econometric model can specify relationships the way they should oper-ate, as our model has tried to do, but the one equation model because of its very simplicity must ignore the problems

Impact on Specific Sectors

A second objection concerns the almost total absence of structural information which comes out of the one equation approach Even if the relationship between monetary variables and GNP were close and predictable, which has not been docu-mented by Friedman-Meiselman, Andersen-Jordan, our model, or any other study, the obligation of monetary authorities, business forecasters, and research econo-

mists would extend beyond knowing what will happen to GNP if we alter

mon-etary policy We want to know how it will happen, when it will happen, and what

we look at to make sure it is happening We must worry about the fate of specifi c sectors—what will happen to unemployment rates and the balance of payments, whether savings and loan institutions will go bankrupt, how many new homes will be built, whether state and local governments will be able to sell securities Our model gives answers to all of these important questions and the one equation approach does not

The econometric problem associated with this structural agnosticism involves the complete avoidance of use of prior information in obtaining statistical estimates As any casual observer of the economics profession knows, lately the computer has fos-tered a tremendous boom in applied econometrics It is now quite easy to put some data together, run them through a statistical program, get results, and write them

up Such an atmosphere encourages lack of caution—caution in making sure you fulfi ll all the assumptions econometrics’ textbooks say are necessary, caution that you have enough data that you really are explaining structure and not simply some few observations, and caution that you have investigated all implications of your results and are sure these results are internally consistent and consistent with any available, prior information Virtually all econometric models leave something to be desired

on these grounds and ours is no exception We have unreliable coeffi cients, unusual feedbacks and inconsistencies, and other defects In fact, suspicion of these defects is what encourages many people to favor simple models in the fi rst place

But the virtues are not all on one side The trouble with a simple model is that

it ignores so much useful structural information For example, we know that GNP

equals consumption plus investment plus government spending Thus if ment spending rises, GNP will rise the same amount unless something else rises

govern-or falls We have govern-orders statistics and are reasonably sure that there are lags in the investment process because we can trace response patterns fi rst through orders and then through fi nal expenditures We are reasonably sure that disposable income aff ects consumption but that the investment credit does not In all of these cases a model can take advantage of the many things we do know about the economy and build them in rigorously—knowledge about identities, lags, the mathematical form

of relationships, and what variables should or should not be included in various equations In all of these cases a one equation approach takes its chances, ignoring prior information and having no checks on internal consistency

We may then want to examine the internal consistency of the Andersen-Jordan one equation model This model says fi rst of all that the response of GNP to money

is greater than the quantity theory would predict, with the whole process ating by the end of four quarters This means that the entire response of con-sumption, investment, housing, unemployment, and prices has been completed within four quarters Short-term rates have responded completely, long-term rates have responded completely, business fi rms and state and local governments have responded completely, the multiplier reaction on consumption is over, these impulses have worked through unemployment rates and raised or lowered prices to their new level, and velocity has overadjusted—all within four quarters

The Andersen-Jordan model shows a coeffi cient on current government ditures substantially less than unity Since government spending is part of GNP, this coeffi cient means that a rise in government spending is accompanied by a large decline in other components of fi nal demand For this to happen, there must be some combination of a very low multiplier and prompt “crowding out” of other expenditures through the fi nancial sector The fi rst possibility implies a very low marginal propensity to consume In fact, the most likely conjecture is that the short- and long-run MPC is negative (or marginal savings rates are in excess of 100%) Such a conjecture would simultaneously explain the weird positive short- and long-run coeffi cients for tax revenues

How Much Elasticity?

The second possible explanation for Andersen-Jordan’s results is the ticity of demand for money or the monetary base If this demand is insensitive to interest rates, fi scal policy can only increase interest rates and “crowd out” other expenditures This explanation fl ies in the face of a countless number of empirical studies showing substantial interest sensitivity of the demand for money and/or near money, including, interestingly enough, the PhD thesis of one Jerry Jordan 4 Which incarnation are we to believe?

I do not mean to carp at the Andersen-Jordan results, which I have already admitted are suggestive and interesting When they imply propositions as startling

as the above, however, disagreeing with new economists and quality theorists alike,

I think Andersen-Jordan have an obligation to tell us more about how their model works Could they, for example, provide other reduced-form equations for interest

rates, consumption, investment, unemployment rates, and prices just so we can low the process, see exactly what is happening, and decide whether we believe it?

A further defect of the Andersen-Jordan approach has already been thoroughly

debated by de Leeuw-Kalchbrenner and Andersen-Jordan in this April’s St Louis

Review, and I will only summarize the issues Just as it is not convincing evidence

that money determines GNP if really GNP is determining money, so it is not convincing that the monetary base determines GNP if GNP really determines the monetary base Thus in any regression of GNP on monetary variables we must make sure the monetary variable is exogenous, for if it is not, one equation results will not tell us any more than that the coeffi cients are meaningless As de Leeuw-Kalchbrenner have shown, if the Fed really takes un-borrowed reserves as exog-enous, the Andersen-Jordan equations give radically diff erent, and to many people much more credible, results Now both taxes and expenditures matter, money mat-ters less, and the lags seem to be much longer In fact, the policy multipliers are now quite close to those of our model Thus our decision as to the fundamental properties of the macro system apparently hinges on what is the exogenous mon-etary variable

If the Fed has regularly used any one monetary variable as a policy target, nobody has discovered what this variable is The more likely possibility is that the Fed has fol-lowed targets on some variables some of the time and other variables other times In addition to the money supply and the monetary base, free reserves and interest rates have been commonly proposed and probably often used as Federal Reserve target variables, a proposition argued even by observers at the Federal Reserve Bank of St Louis 5 If either one of these variables really were the target, moreover, it would be closer to the truth to use unborrowed reserves as the exogenous monetary variable, thus favoring the de Leeuw-Kalchbrenner view The Andersen-Jordan response that you can only use one side of the base identity misses the point completely—you should use whatever side the Fed uses—as does their regression attempting to show that the central bank off sets borrowed reserves We get a perfectly satisfactory answer to that regression just by looking at bank behavior Whatever one thinks

is the exogenous monetary variable, therefore, we would all agree that to say the basic properties of the macro system depend on what is exogenous, which we do not know, is a very unsatisfactory way to leave the question In the last analysis the Andersen-Jordan regressions have raised more questions than they have answered

In fact, if the 1965 pattern repeats its-self, this will stimulate somebody else to build

an econometric model to get to the bottom of these issues

The discussion thus far has centered on what the two models can tell us about monetary policy and the way it infl uences economic activity But since the topic of this conference is forecasting, we still may want to ask how the two models com-pare as forecasting devices

Forecast Errors

The one quarter standard error for the FRB-MIT econometric model is $2.4 billion for current dollar GNP This error gradually increases as we try to forecast more into the future We can predict GNP two quarters ahead with a standard error

of $4.0 billion and four quarters ahead with an error of $7.1 billion After that, due

to the inherent stabilizing properties of the model, our standard error of forecast stays in the $8 to $9 billion range for two-three years and is only about $15 billion for periods as long as ten years

We can compare this with the one quarter forecast errors produced by Jordan, which average about $4.1 billion for one quarter ahead We notice that these errors are larger than those of the FRB-MIT model for two quarters ahead

Andersen-In other words, our model can in a sense see twice as far ahead with the same clarity Andersen-Jordan have provided no forecast errors further ahead than one period, but since their model has no obvious stabilizing properties, I think we would expect it to drift off much more rapidly than does ours and be substantially worse in longer-run forecasting

If you don’t care about long-run forecasting, you may inquire whether it is worth the eff ort to add ninety-nine equations merely to reduce the one quarter forecast error by $1.7 billion My answer is to turn the tables on Andersen-Jordan and confront them with a model even simpler than theirs which works almost as well in short-term forecasting Let me propose a model where current dollar GNP grows by exactly 1.42 percent per quarter, quarter in, quarter out This model is not as good in business cycles, but over the 1952–1968 period it would nevertheless have predicted current dollar GNP one quarter ahead with

a standard error of only $5.4 billion, closer to the Andersen-Jordan one quarter standard error than theirs is to that of the FRB-MIT model Thus if we want real simplicity, let’s go to my 1.42 percent model It doesn’t even have any variables

We fi nd that monetary policy is quite important, but by no means the only force infl uencing aggregate demand Fiscal policy also matters, as do exports, population, and a host of other exogenous variables In addition to having a powerful infl uence, monetary forces are found to operate with a long and variable lag, which at least raises the possibility that using short, constant lags will be inadequate

We are confronted with a simple one equation model which gives the ance of working very well It says that only money matters, that money matters

appear-by twice as much as the quantity theory would have predicted, with very short lags, and that fi scal policy does not matter at all These conclusions are suffi ciently alarming that they should be investigated, especially when the conclusions do not stand up to diff erent yet entirely reasonable assumptions as to what is exogenous And on inspection, we fi nd that one equation may not forecast so well either For

one quarter ahead it beats a very naive model by only $1.3 billion and loses to our model by $1.7 billion

I do not deny that econometric models have problems This is still a very early stage in their development, they are complicated and costly, and their implications are not easily understood The last year has been a chastening experience for all models, ours included, even though for many years before this they have worked tolerably well Thus while I admit that our econometric model is complicated, costly, and presently leaves much to be desired, I think you should be quite wary before succumbing to the wiles of our low-cost competitor It may very well be that monetary forces are complicated enough to defy even simple models

Acknowledgments

The views expressed herein are my own and do not refl ect those of the Federal Reserve Board I thank John Kalchbrenner, Frank de Leeuw, and Harvey Galper for their helpful comments on this chapter

Notes

Originally published in Business Economics, Vol 4, No 4 (September 1969), pp 21–26

1 There are several write-ups of our model De Leeuw and I have published articles

in the Federal Reserve Bulletin for January, 1968, and June, 1969, Harold Shapiro and Robert Rasche have an article in the May, 1969, American Economic Review, and Ando and Modigliani have one in the May, 1969, American Economic Review

2 Andersen-Jordan assented to the price adjustment after a comment by de Leeuw and John Kalchbrenner

3 Andersen-Jordan did try to answer to these defects by using the full employment budget surplus weighted for differential demand impacts In fact, they used weights provided by me But even though it is better to use a weighted surplus than an unweighted surplus, there are many defects with a weighted surplus as compared with

an econometric model The values of the weights depend on how the model works, relevant time dimensions, and, for taxes, whether reductions or increases are brought about by rate changes or incentive features—all questions which cannot be solved satisfactorily no matter what weights are used

4 See J L Jordan, “The Market for Deposit-Type Financial Assets,” UCLA PhD thesis,

1969

5 An article by Michael Keran and Christopher Babb, two colleagues of Andersen and

Jordan, in the June, 1969, St Louis Review argues at length that free reserves is “the

most reliable indicator of monetary policy” (John Wood’s phase)

1969

ECONOMETRIC MODEL BUILDING FOR

GROWTH PROJECTIONS

Lawrence R Klein, University of Pennsylvania

Emphasis in much recent work on macroeconometric model building has been placed on short-term forecasting This is by no means a misguided activity, but it leaves undone another task of great importance to the users of econometric output There is an evident need for longer-term analysis, capa-ble of providing projections ahead for a decade or more Just as many users find

it essential to look ahead for the next few months or for periods up to two or three years, others find it essential, for their purposes, to look ahead for much longer stretches of time

It is reasonable to ask whether short-term models might in fact be used for term analysis, as well, by simply projecting them several short periods into the future

longer-It would be a great economy of eff ort if we could, indeed, use one overall model simultaneously for short and long-term analysis Attempts at extrapolating a short-term model into the future for a decade or more is one approach to the problem of growth projection, and this lead will be considered In following this lead, however, defi ciencies in the approach will become evident, and we shall be led to consideration

of an alternative research strategy, namely the idea of building a specialized long-term model which is better designed for applicability to growth projections 1

Normally, the Wharton Model is extrapolated ahead every quarter for 8 ters On occasion we have extrapolated for 10–12 quarters, but it is only recently that we have considered long-range extrapolations of 5 or more years, up to a max-imum of 25

Long-Term Extrapolation of Short-Term Models

Given the budgetary practices of government, a forecast ahead for one year (by smaller time units preferably) covers a sensible period for which it is possible to deal

with econometric forecasting on a scientifi c basis In the case of Wharton Model forecasts, we feel comfortable with the projection of the fi rst four quarters It is necessary, however to look ahead even further in order to gain an impression of the

economy’s expected time path This accounts, roughly, for our interest in two-year,

of the work week The particular six-year projection is designed so that ment is held near to 3–4 percent over the course of the next six years Similarly, short-term interest rates are not allowed to fall below 4.5 percent

Any forecast for a period as far ahead as fi ve to six years must make some cifi c assumptions about the termination of the war in Viet Nam, demobilization, and a return to more of a peace time economy Hopefully, these considerations should be relevant to short-run forecasts of less than two years’ duration In the six-year extrapolation of the Wharton Model, it is assumed that a peace settlement

spe-is reached by January 1, 1970, and that the fi rst quarter of that year spe-is the ning of a modest decline in government military expenditures and demobilization

begin-On a gradual basis, defense spending is assumed to be reduced by $5.0 billion in

Wharton-EFU model, selected variables

1958 prices, while the number of persons under arms is assumed to fall by 340,000 persons Although the assumed cut in defense spending appears to be small, it is accompanied by a substantial demobilization This changed “mixture” in compo-sition of the military establishment is meant to refl ect the view that offi cers will prevail on the civilian government to provide them with new hardware and R &

D for further developments These requests follow a period of preoccupation with

an irritating war

Fiscal Assumptions

To compensate for the reduced amount of military activity, taxes are assumed to be cut by elimination of the surcharge—a reduction to 5 percent on January 1, 1970, and complete elimination by July 1, 1970 This is the present Administration posi-tion, apart from the war situation We have also programmed elimination of the investment tax credit from April, 1969 Although military spending is programmed

to be cut after the fi rst quarter of 1970, total government expenditures are assumed not to fall The growth in state and local expenditures plus added civilian expendi-tures at the federal level keep the constant dollar total rising after the cessation of hostilities, fi rst by $0.2 billion per quarter (annual rate), then by $0.4 billion, and then by $0.6 billion In addition, there is an assumption that normal growth will continue in social security contributions and in transfer payments

Population trends are assumed to continue over the six years; world trade is projected on its course of secular growth; and monetary policy is assumed to be somewhat eased after the fi nancial burden of the war is lessened Even so, interest rates stay at a level as high as 4.5 percent (short-term) or 5.5 percent (long-term) The outcome of this intermediate length projection is a hesitating phase of demobilization and transition to peace; whereupon the economy quickly regains its regular growth trajectory It is a well balanced growth, but all prices do not rise uniformly Prices related to services grow steadily (20% over six years), but some

Six-year projection (billions of dollars)

1971.4 1972.1 1972.2 1972.3 1972.4 1973.1 1973.2 1973.3 1973.4 1974.1 1974.2 1974.3 1974.4

660.9 670.8 681.1 690.0 701.6 712.7 723.9 733.8 746.3 758.2 770.1 780.0 792.9 110.1 112.4 114.8 116.5 117.9 120.1 122.7 124.5 126.0 128.3 131.2 133.3 134.8 43.4 44.2 44.6 46.2 47.4 48.2 48.6 50.3 51.5 52.3 52.9 54.6 55.9

239.7 243.6 247.4 251.1 254.8 258.6 262.3 266.1 270.0 274.0 277.9 281.9 285.9 1,064.9 1,081.3 1,099.1 1,114.2 1,133.2 1,151.5 1,169.7 1,186.5 1,205.8 1,225.3 1,245.7 1,261.5 1,282.0 778.9 785.4 791.6 797.3 805.4 812.8 818.4 824.3 831.7 839.4 846.3 851.7 859.8 136.7 137.7 138.8 139.7 140.7 141.7 142.9 143.9 145.0 146.0 147.2 148.1 149.1 105.3 105.8 109.3 112.2 113.3 113.4 116.8 120.2 120.9 120.8 124.5 127.3 128.0

durable goods’ prices rise much more slowly (13% over six years) The peace pause lasts for less than two years and is not expected to develop into a recession such as the one we had after the Korean War

It is interesting to note that if the economy can continue to grow at full ment for a number of years, tax revenues come in quite strongly, while transfer payments are restrained; therefore the budget defi cit withers away It builds up

employ-in the transition period and then declemploy-ines If the solutions were projected at full employment for a few more years, budget balance and a surplus would gradually

be reached We have noticed this result in 15–25 year simulation projections of the Model This is an eventual “fi scal dividend.”

In order to project the Wharton Model ahead for six years, it was necessary

to make strong assumptions about the growth of the self-employed, government employment, productivity, and labor force These exogenous assumptions have a strong infl uence on the unemployment rate Also, the equations for nonmanufac-turing hours worked per week have strong downward trends This variable tends to get far out of line in connection with rising wage rates in the solution To keep the solution pattern in balance for long periods, it is necessary to make frequent adjust-ments to some of the equations; otherwise their solutions drift over 24 quarters This is not much of a problem in one or two-year extrapolations, which are the primary objective of a system like the Wharton Model

A special assumption of this programmed solution is that residential tion (including mobile homes, recorded under consumer outlays on automobiles and parts) will be a strong trend growth sector, reaching price-corrected levels in excess of current outlays by about 40 percent in six years

This particular model has been solved dynamically for 15- and 25-year periods ahead These solutions have been generated under both deterministic and stochas-tic conditions A short-term forecasting model may be useful in giving longer-run information, but it must be handled very carefully and adjusted, so as to keep it on track It is a useful start, but is only an expedient substitute for a genuine long-run model

Specification and Characteristics of a Long-Run Model

A more satisfactory approach to the problem of long-range forecasting is to build

a fresh model specifi cally designed to exhibit the processes of economic growth

A self-evident prerequisite for such a research project would be the tion of a data base that, itself, incorporates a long span of economic growth Some parameters of simple models that are known to be capable of generating dynamic growth trends could conceivably be estimated from short samples of recent data Also, recent cross sections samples could be used to estimate some typical parameters of a growth model, for it is often argued that cross section based relationships come closer to picturing long-run than short-run patterns But without a long-run data base, there is limited possibility for testing the accu-racy of growth models by their ability to interpret observed trends In addition, some socio-economic processes that tend to get overlooked in short-run business cycle analysis will not show up in recent short-run data samples, and must be

prepara-based on observations from a longer time span In general, we shall be shifting variables from an exogenous (short-run) category to an endogenous (long-run) category in building a growth model

To be quite pragmatic, a growth model should be estimated at least on a sample that goes back as far as 1929, only because the offi cial national income and prod-uct accounts of the US Department of Commerce begin from that date A sample period beginning in 1919 would, in many respects, be better, but there are diffi -culties in reconstructing a complete set of national accounts for the 1920s 2 In any event, whether the sample period begins with 1919 or 1929, a long range study designed for growth projections should be based on annual data It is extremely diffi cult, but not impossible, to extend quarterly series as far back as the 1930s and 1920s, but one should not be pre-occupied with very short-run fl uctuations, espe-cially those associated with inventory cycles, in such analysis, and annual data would seem to be better suited to the task

For some purposes, it is possible to construct annual data series as far back as

1890 or 1900, but these would mainly serve small models and be limited in the types of variables that could be projected 3 On a somewhat rougher plane, decade accounts can be estimated for the period since the Civil War Rudimentary models can be estimated from these long-run statistics 4 The decade data smooth out more than short-run inventory cycles; they eliminate the classical business cycle

It has been remarked that a long-run model ought to “endogenize” a number of variables that are now placed in the exogenous category These are, especially, vari-ables connected with government fi scal and monetary policies and demographic magnitudes In some short-run econometric models, the expenditure and fi nancial decisions of state and local governments are placed partly in the endogenous cat-egory This is all the more compelling for a long-run model The principal expen-diture magnitudes for state and local governments are for education and highways These, in turn, depend on the human and automobile populations, both of which should be explained by the model The automobile stock is generated by con-sumer expenditures on cars and a depreciation relationship These are both endog-enous processes The whole matter of modeling demography is a major issue to be explained below

As far as state and local revenues are concerned, they are generated by the income and asset bases of the population, given the tax laws Financial decisions on the issuance of state and local securities are tied to interest rates and other money market phenomena determined in the monetary part of a model

Federal Spending

Federal expenditures are of at least three basic types: (1) national defense; (2) discretionary civilian expenditures; (3) non-discretionary civilian expendi-tures For the most part, defense expenditures are exogenous They have changed

in character enormously in this century and have been determined under quite diff erent criteria in each of the post-war governments from Presidents Truman

to Nixon It seems best to leave these in the exogenous category for long-term

analysis Non-defense expenditures can be made according to specifi c programs that any administration decides (with Congressional support) to carry out These discretionary expenditures, such as crop support programs, Medicare, social insurance extension, public works, and the like are also exogenous But much

of government outlay is associated with population growth and the normal development of fi lling the country’s needs These should be explained by pop-ulation and income variables Taxes and many transfer payments will be related

to income and activity bases according to the prevailing laws Short-run models already contain such relationships, and it is nothing new to include them in longer-run models

The “endogenized” government sector depends signifi cantly on demographic characteristics, as do other economic processes, such as house construction, con-sumer expenditures, and labor force development A major part of the widening

of the endogenous sector of a model for long-run analysis will be associated with population growth in its various dimensions

In the fi rst place, total population must be projected Following the original gestion of Valavanis-Vail, we can form the identity

N t = N t − 1 (1 + b t − d t + i t )

N t = population at t

b t = birth rate

d t = death rate

i t = net immigration rate

The econometric problem, therefore, is to explain birth, death, and immigration rates These are likely to depend on rates of income growth, international income diff erentials, and trends

Total population projections are not, however, adequate for the growth lem The age composition is signifi cant for education demands; the age and sex composition is needed for determining labor force participation rates; the age, sex, and race composition for unemployment and its eff ects on wage rates; and fi nally marriage or household formation rates for housing demand Some of these demo-graphic variables are largely exogenous, while others have an appreciable element

prob-of endogeneity

Econometricians have largely neglected careful consideration of demographic processes, but they are in an excellent position to interrelate social and economic factors that determine population dynamics 5

The distinctive features of a long-run growth model may show up in ways other than the transference of variables from the exogenous to the endogenous category, together with the addition of new explanatory equations Some variables and types

of equations are likely to be modifi ed as we focus attention from the short to the long run Inventory equations may be consolidated with fi xed investment, or if left as separate equations, they may be simple transaction relationships between stock holding and production In wage determination equations, the infl uence of short-run labor market conditions (unemployment ratio) may be downgraded, and

the relationship of wage changes to productivity changes may dominate the wage equation in the long run Other dimensions of unemployment, such as its demo-graphic composition, may be more important than the overall level of unemploy-ment in the long run

In the original mathematical formulations of the Keynesian theory of employment, it was artifi cially declared that a short-run period was being con-sidered in which the stock of capital could be taken as given On this basis, the production function of the model was expressed as a simple bivariate rela-tionship between labor input and total output This is surely an unsatisfactory specifi cation of the model, even for the short run, because nonzero investment

is capable of being estimated from the propensity-to-invest function, and this could not be consistent with a given and fi xed stock of capital In the short or the long run, it is essential to have both labor and capital inputs in the aggregate production function It is plausible to consider a concession to the idea of the use of a capital variable for short-run analysis by modifying the variable from K

to uK, where u is an index of the rate of utilization of capital facilities In the long run, we can smooth the typically cyclical variable u, by fi xing it at some constant level, either unity or just below unity The production function in a long-run model should therefore depend on both labor and capital as inputs, and if it is at all possible, the capital variable should be corrected for quality change Instead of the usual measure,

K t = stock of capital at end of period t

I t = gross investment of period t

D t = capital consumption of period t

we can use the “vintage” measure,

These measures weight investment i periods ago with the coeffi cient V (i) or λ i The

“vintage” type measure is one possible device for taking care of technical change, but a productivity trend, specifi c if available, should be considered as an additional variable for the long-run production function